The possibly unsolvable problem of polls
Polls are a problem for which there may be no solution.
Today is Iowa caucus day. Pollsters have been working double shifts for weeks. Will Obama pull ahead of Clinton? Will somebody--anybody--pull ahead in the Republican race? As of last night, results of the Reuters/C-SPAN/Zogby poll predict that today the results will be Obama-Edwards-Clinton for the Democrats and Huckabee-Romney for the Democrats.
What's the public value of poll results? I understand the value to candidates who appear to be in the lead; it's just one more way for them to say to the rest, "Give it up. The people want me," months before the election. And I understand the value to candidates who appear to be behind; it alerts them to look for ways they can drastically change their core principles in order to be more appealing.
But it seems to me that the effect of poll results on the general public is merely to skew the next poll result and, ultimately, the election. People leaning toward the candidate polling highest will feel vindicated; people leaning toward a candidate polling low will question their choice (who likes to waste a vote on a loser?); and the undecideds will find themselves liking the poll winners more and more, even if they consciously fight the impulse. (Can thousands of people be wrong?)
Polls skew election results. We're only human. We react to information.
What to do? Publishing poll results will never be outlawed: after all, this is America, and if the First Amendment to the Constitution gives us the "right" to Internet porn and prime-time erectile dysfunction commercials, it certainly also gives us the right to know poll results. Only a voting citizenry that elects officeholders based on their past actions rather than the things they say on-camera will be immune to the lemming call of poll results.
Which is why I'm skeptical about solving this particular problem.
(flickr photo by enfocar1200)








